Skip to main content

​
Purpose

Three separate capability questions about one country: who is of working age and for how much longer, what the education pipeline is producing, and whether the workforce holds the industrial skills a manufacturing base needs. Served from GET /api/resilience/v1/get-demographics-capability?countryCode=<ISO2> and MCP get_demographics_capability. Both require a Pro subscription.

​
Three independent groups

The response is deliberately not a single blended index. Each group resolves on its own, from its own source, on its own clock.

​
Age structure — UN World Population Prospects

medianAgeYears, workingAgePopulationPeople, oldAgeDependencyRatioPercent, totalDependencyRatioPercent, and a forward workingAgePopulationProjected10yPeople. WPP publishes revisions every two years; between revisions the projection is the same series, not a fresh estimate.

​
Education pipeline — UNESCO UIS and World Bank WDI

tertiaryEnrollmentGrossPercent, stemGraduatesSharePercent, and researchersPerMillion. UIS is primary; WDI fills where UIS coverage lapses. Reporting lag is long — three to five years is ordinary, not a fault.

​
Industrial workforce — ILOSTAT

manufacturingEmploymentSharePercent, trainedIndustrialWorkforcePeople, craftTradesEmploymentPeople, and plantMachineOperatorsEmploymentPeople. ILOSTAT coverage is the thinnest of the three: a country can have current WPP age structure and no usable ILOSTAT observation at all.

​
Reading the response

Every metric carries four things: the value, its year, its source, its unit, and an explicit available flag. Read available before the value. These are proto3 messages — an absent number serializes as 0, and a zero median age or a zero manufacturing share is never a real observation. The flag is the only thing separating “measured” from “missing”. Do not average across groups. They have different vintages and different coverage. A country whose ILOSTAT row is eight years old and whose WPP row is current does not have an eight-year-old demographic profile; it has two observations that must be read separately. The response keeps them apart for that reason. Retention lives on stages[].status, not on each value. A stage is fresh, retained, or unavailable. retained means that source group was carried forward from a previous fetch because the current upstream request failed. Individual observations have no retained field. The top-level available: false means no group resolved at all — the country has no usable observation in any of the three, which is different from a country whose groups each carry their own partial availability.

​
Cadence

The seeder refreshes on the slowest useful cycle for the set. Upstream publication is the binding constraint: WPP every two years, UIS and WDI annually with multi-year lag, ILOSTAT irregularly by country.